STEMPO - dynamic X-ray tomography phantom

The Spatio-TEmporal Motor-Powered (STEMPO) phantom is a physical phantom designed for collecting dynamic X-ray tomography data. The dynamic part of the phantom is computer controlled allowing for wide variety of different measurements and sampling setups to be used. The primary goal is to help mathematical community test and validate novel dynamic tomography reconstruction methods. Detailed documentation of the phantom, the included data (volume 1 only) and some examples can be found on the related publication: https://doi.org/10.1007/978-981-97-6769-4_1 (available as an arXiv preprint: http://arxiv.org/abs/2209.12471). This data set can be appended with new data in the future. Current version (1.2) includes Data - vol.1 (v1.0) stempo_static_2d_b*.mat stempo_static_3d_b*.mat stempo_cont360_2d_b*.mat stempo_cont360_3d_b*.mat stempo_seq8x45_2d_b*.mat stempo_seq8x45_3d_b*.mat stempo_data_geometries.csv Data - vol.2 (added in v1.2) stempo_seq8x180_2d_b*.mat stempo_seq8x180_3d_b*.mat where b* denotes downsampling or binning of the data by a factor of (4, 8, 16 or 32). These are 2D and 3D data collected from a static object for reference, or from a dynamic target in a continuous 360 projection scan or sequence of 8 rotations, each consisting of 45 or 180 projections (with seq8x45 and seq8x180 data respectively). Finally stempo_data_geometries.csv is a simple table containing the key parameters of the measurement geometry in text format. Note that the height of the phantom for volume 2 data is slightly different compared to volume 1, including the static scan (mostly relevant for comparing 3D reconstructions). In addition the data set contains Additional files stempo_ground_truth_2d_b4.mat which is an approximation of the true motion obtained from a single static FBP reconstruction which has been interpolated to match the location of the moving block during the cont360 and seq8x45 scans. Finally there are Example algorithms stempo_fbp_example.m stempo_fdk_example.m stempo_pdfp_wavelet_2d_example.m stempo_LplusS_2d_example.m which are short example algorithms of well know analytic (FBP and FDK) and iterative methods. stempo_pdfp_wavelet_2d.m uses variational regularization and wavelet transform of the 2D + time object to reach a suitable solution. The codes are adapted from [1,2]. stempo_LplusS_2d_example.m attempts to split the reconstruction into low-rank component L and a sparse dynamic component S. This code is adapted from [3]. These are meant to give users ideas how the data can be used in different applications to match the requirements of different methods. Easiest way to utilize the data is with the ASTRA Toolbox and the HelTomo Toolbox. Some of the example codes also require Spot Linear Operator Toolbox (highly recommended) and the Wavelet Toolbox. However none of these are mandatory and any method (including programming languages other than MATLAB) are fine as long as the measurement geometry is respected. The author is supported by the Emil Aaltonen Foundation junior researcher grant no. 200029 and the Vilho, Yrjö and Kalle Väisälä Foundation of the Finnish Academy of Science and Letters. The author also acknowledges the support of Academy of Finland through the Finnish Centre of Excellence in Inverse Modelling and Imaging 2018–2025, decision number 312339. Finally the author would like to thank E. Heikkilä, T. Heikkilä, A. Meaney and F.S. Moura for all their technical expertise and help in developing, building and imaging the mechanism. The author also thanks O. Tapaninen for helping measure the data for vol.2.时空电机驱动（Spatio-TEmporal Motor-Powered, STEMPO）体模是一款专为采集动态X射线断层扫描数据设计的物理体模。该体模的动态部分由计算机控制，可适配多种不同的测量与采样配置方案，其核心目标是协助数学领域研究者测试并验证新型动态断层重建算法。 有关该体模、所含数据（仅第一卷）及部分示例的详细文档，可参阅相关出版物：https://doi.org/10.1007/978-981-97-6769-4_1（亦可作为arXiv预印本获取：http://arxiv.org/abs/2209.12471）。 本数据集未来可追加新增数据。当前版本（v1.2）包含： - 数据卷1（v1.0） stempo_static_2d_b*.mat stempo_static_3d_b*.mat stempo_cont360_2d_b*.mat stempo_cont360_3d_b*.mat stempo_seq8x45_2d_b*.mat stempo_seq8x45_3d_b*.mat stempo_data_geometries.csv - 数据卷2（v1.2版本新增） stempo_seq8x180_2d_b*.mat stempo_seq8x180_3d_b*.mat 其中b*代表将数据以4、8、16或32的倍率进行下采样或分箱处理。上述数据包含从静态参考物体采集的二维与三维数据，或是从动态目标采集的、在连续360°投影扫描下或8次旋转序列下的数据：每次旋转分别包含45或180次投影（对应seq8x45与seq8x180数据集）。此外，stempo_data_geometries.csv为以文本格式存储的简易表格，包含测量几何的关键参数。请注意，第二卷数据所用体模的高度与第一卷略有差异，其中也包含静态扫描数据（该差异主要用于三维重建的对比研究）。 此外，本数据集还包含以下附加文件： stempo_ground_truth_2d_b4.mat 该文件为真实运动的近似值，由单次静态滤波反投影（Filtered Back Projection, FBP）重建得到，并经过插值以匹配cont360与seq8x45扫描过程中移动模块的位置。 最后，本数据集还提供以下示例算法： stempo_fbp_example.m stempo_fdk_example.m stempo_pdfp_wavelet_2d_example.m stempo_LplusS_2d_example.m 上述代码均为针对经典解析算法（FBP与锥束滤波反投影（Feldkamp-Davis-Kress, FDK））及迭代算法的简易示例。其中stempo_pdfp_wavelet_2d.m采用变分正则化与二维+时间对象的小波变换，以得到合适的重建结果，该代码改编自文献[1,2]。stempo_LplusS_2d_example.m旨在将重建结果分解为低秩分量L与稀疏动态分量S，该代码改编自文献[3]。这些示例代码旨在为使用者提供思路，以根据不同算法的需求，将数据集应用于各类场景中。 使用本数据集的最简方式为借助ASTRA工具箱与HelTomo工具箱。部分示例代码还需依赖Spot线性算子工具箱（强烈推荐）与小波工具箱，但上述工具均非强制依赖项，只要遵循测量几何规则，使用者可采用任意方法（包括MATLAB之外的编程语言）进行处理。 本研究得到埃米尔·阿尔托宁基金会（Emil Aaltonen Foundation）青年研究者基金（编号200029）以及芬兰科学与文学研究院旗下的维尔霍、于尔约与卡莱·魏斯莱基金会（Vilho, Yrjö and Kalle Väisälä Foundation）的资助。作者同时感谢芬兰科学院通过2018-2025年芬兰反演建模与成像卓越中心项目（项目编号312339）提供的支持。最后，作者感谢E. Heikkilä、T. Heikkilä、A. Meaney与F.S. Moura在该装置的研发、搭建与成像过程中提供的专业技术支持与帮助。 作者同时感谢O. Tapaninen协助采集第二卷的数据。